Skip to content
TrustList
AI Models

AI Model Rankings

Models are ranked by the Artificial Analysis Intelligence Index v5.0.1 wherever it has tested them. Models it has not tested follow, ordered by a blend of public benchmarks, verified reviews and community votes. Every model links to its full profile.

Vote#ModelBlendCapabilityRatingArtificial Analysis Intelligence Index v5.0.1ARC-AGI-3 (Provider Adapter harness)ARC-AGI-3 (Standard harness)Artificial Analysis Intelligence IndexArtificial Analysis Intelligence Index v4.3.2ArXivMath
1
Google · 1,000,000 ctx · Proprietary
60
+8 vs avg
60-60 index
+23.78
-----
2
Anthropic · 1,000,000 ctx · Proprietary
59
+7 vs avg
59-59 index
+22.78
---58 index
+9.6
-
3
OpenAI · 1,050,000 ctx · Proprietary
82
+30 vs avg
82-57 index
+20.78
99.9 %62.7 %---
4
Anthropic · 1,000,000 ctx · Proprietary
56
+4 vs avg
56-55 index
+18.78
---56 index
+7.6
-
5
OpenAI · 1,050,000 ctx · Proprietary
54
+2 vs avg
54-55 index
+18.78
---52 index
+3.6
-
6
OpenAI · 1,050,000 ctx · Proprietary
51
−1 vs avg
51-51 index
+14.78
-----
7
Google · 1,000,000 ctx · Proprietary
52
+0 vs avg
52-50 index
+13.78
---53 index
+4.6
-
8
Anthropic · 1,000,000 ctx · Proprietary
52
+1 vs avg
52-49 index
+12.78
-----
9
Moonshot AI · 3T-class (per Moonshot) params · 1,048,576 ctx · Weights promised by 4 November 2026; licence not yet named
46
−6 vs avg
46-46 index
+9.78
-----
10
Meta · 1,048,576 ctx · Proprietary (no licence stated; weights not published)
44
−8 vs avg
44-44 index
+7.78
-----
11
Anthropic · 1,000,000 ctx · Proprietary
36
−16 vs avg
36-36 index
−0.22
-----
12
xAI · 500,000 ctx · Proprietary
41
−11 vs avg
41-35 index
−1.22
---46 index
−2.4
-
13
Alibaba · 2.4T (95B active) params · 1,000,000 ctx · Proprietary (weights announced)
60
+9 vs avg
60-34 index
−2.22
-----
14
Z.ai (Zhipu AI) · 1,000,000 ctx · Proprietary at launch; weights released ~2 weeks later
58
+7 vs avg
58-32 index
−4.22
-----
15
Google · Proprietary
59
+7 vs avg
59-31 index
−5.22
-----
16
Xiaomi · 1.02T total, 42B active (sparse mixture of experts) params · 1,000,000 ctx · MIT — open weights (published as MiMo-V2.6-Pro-RL)
38
−14 vs avg
38-30 index
−6.22
---46 index
−2.4
-
17
StepFun · 600B total, 27B active per token (sparse mixture of experts) params · 1,000,000 ctx · Proprietary — BF16 open weights announced for 15 October 2026, not published yet
37
−15 vs avg
37-29 index
−7.22
---44 index
−4.4
-
18
MiniMax · Proprietary (weights to be released later, per MiniMax)
29
−23 vs avg
29-29 index
−7.22
-----
19
OpenAI · 1,050,000 ctx · Proprietary
33
−19 vs avg
33-28 index
−8.22
---37 index
−11.4
-
20
Z.ai (Zhipu AI) · 1,000,000 ctx · Open weights
28
−24 vs avg
28-28 index
−8.22
-----
21
Xiaomi · 309B total, 15B active (sparse mixture of experts) params · 1,000,000 ctx · MIT — open weights (published as MiMo-V2.6-Flash-RL)
27
−25 vs avg
27-27 index
−9.22
-----
22
Mistral AI · about 1T total (1.05T per Mistral docs), 52B active, plus 1.6B vision encoder params · 1,048,576 ctx · Mistral Ecosystem Open-Weight License (modified MIT; model-as-a-service providers need a commercial licence)
26
−26 vs avg
26-26 index
−10.22
-----
23
Meta · 816B total parameters, 42B active per token (mixture of experts, 256 experts with 8 active per token) params · 1,000,000 ctx · Muse Model License
25
−27 vs avg
25-25 index
−11.22
-----
24
DeepSeek · 1,000,000 ctx · Proprietary
59
+7 vs avg
59-23 index
−13.22
-----
25
Alibaba · 27B params · 1,000,000 ctx · Qwen (open weights)
20
−32 vs avg
20-20 index
−16.22
-----
26
NVIDIA · 120B total parameters, 12B active (hybrid Mamba-2, mixture of experts and attention layers, with a C-RADIO vision encoder) params · 1,000,000 ctx · OpenMDW-1.1
14
−38 vs avg
14-14 index
−22.22
-----
27
IBM · 8B params · 512,000 ctx · Apache 2.0
26
−25 vs avg
26-5 index
−31.22
-----
28
Anthropic · 1,000,000 ctx · Proprietary
96
+44 vs avg
96-------
29
Anthropic · 1,000,000 ctx · Proprietary
89
+37 vs avg
89-------
30
Shanghai Artificial Intelligence Laboratory · 744B total, mixture of experts, built on GLM-5.2 (active parameters not published) params · 256,000 ctx · MIT
87
+35 vs avg
87-------
31
OpenAI · 200,000 ctx · Proprietary
84
+32 vs avg
80-------
32
Google · 1,000,000 ctx · Proprietary
80
+28 vs avg
77-------
33
Unbiased AI · 262,144 ctx · Proprietary
68
+16 vs avg
68------88 %
34
Google · 1,000,000 ctx · Proprietary
64
+12 vs avg
64-------
35
OpenAI · 128,000 ctx · Proprietary
64
+12 vs avg
61-------
3661
+9 vs avg
61-------
37
IBM · 30B params · 512,000 ctx · Apache 2.0
57
+5 vs avg
57-------
38
xAI · 500,000 ctx · Proprietary
53
+1 vs avg
53----61 index44 index
−4.4
-
3949
−2 vs avg
61-------
40
OpenAI · 1,050,000 ctx · Proprietary
48
−4 vs avg
48-----48 index
−0.4
-
41
Google · Restricted — Fairwind Program
47
−4 vs avg
47-------
420--------
43
Agnes AI · 33B (open-weights Preview checkpoint; API model size not published) params · Apache 2.0 (Preview checkpoint weights)
0--------
44
Aikido Security · 504B total, ~40B active params · GLM-5.3 License — open weights
0--------
45
Cantina Security (with Yeta) · 321B total, 18B active (mixture of experts) params · 1,048,576 ctx · MIT
0--------
46
Ai2 · 8B params · Apache 2.0
0--------
47
PrismML · 27B, compressed to ternary weights: 1.76 effective bits per weight and a 5.9 GB footprint params · 262,144 ctx · Apache 2.0
0--------
48
Anthropic · 1,000,000 ctx · Proprietary, API only
0--------
49
Anthropic · Proprietary
0--------
50
Anthropic · 1,000,000 ctx · Proprietary, access by application
0--------
51
Cloudflare · 27B params · 64,000 ctx · Apache 2.0
0--------
52
Cloudflare · 9B params · 24,000 ctx · Apache 2.0
0--------
53
Cloudflare · 30B total, 3B active (mixture of experts) params · Apache 2.0
0--------
54
Contrastive-LM · 8B (frozen Qwen3-8B encoder with two small trained heads) params · Apache-2.0
0--------
550--------
560--------
57
Cua · 706,048 trainable parameters — a 2.8 MB checkpoint params · MIT — weights, training code and dataset published
0--------
58
Liquid AI · 3.12B params · 32,768 ctx · LFM Open License v1.0
0--------
59
Liquid AI · 587M params · 16,384 ctx · LFM Open License v1.0
0--------
60
VIDRAFT · 180B mixture of experts (10 of 512 routed experts active per token, plus a shared expert); a fine-tune of Qwen3.8-Flash-Next params · 262,144 ctx · Qwen Community License 1.0
0--------
61
CloudRaker · Proprietary
0--------
62
DeepSeek · 1,000,000 ctx
0--------
63
Nace AI · about 9B (8,953,803,264, BF16) params · 16,384 ctx · Nace.AI Open RAIL-M (modified Open RAIL-M; commercial use above set thresholds needs a separate licence)
0--------
64
ElevenLabs · Proprietary
0--------
65
ElevenLabs · Proprietary
0--------
66
Cohere · 128,000 ctx · Proprietary
0--------
67
Cohere · 128,000 ctx · Proprietary
0--------
68
Google · 740M (270M text-only core) params · 8,000 ctx · Apache 2.0
0--------
69
Fireworks AI · 2.78T (mixture of experts; the Kimi K3 base) params · 1,048,576 ctx · Proprietary
0--------
70
Baidu · 280B total, 14B active (sparse mixture of experts) params · 1,000,000 ctx · ERNIE 5.5 Flash Base Model License Agreement v1.0 (Baidu's own licence, open weights)
0--------
71
Technology Innovation Institute · 1.6B params
0--------
72
Black Forest Labs · 7B params · FLUX Kommunity License v1.0
0--------
73
Black Forest Labs · Commercial weights licence
0--------
74
Black Forest Labs · 5.6B parameters (image generator; text encoder not counted) params · FLUX Kommunity License (open weights)
0--------
750--------
76
Google · 8,192 ctx · Proprietary
0--------
77
Google · 8,192 ctx · Proprietary
0--------
78
Google · 131,072 ctx · Proprietary
0--------
79
Google · 131,072 ctx · Proprietary
0--------
80
Google · Proprietary, API only
0--------
81
Google · 131,072 ctx · Proprietary, API only
0--------
82
Elorian · Proprietary (API and chat UI); post-trained on an open-weights base model
0--------
83
Fastino · 340M (developer’s model card; the models page says 355M) params · Apache-2.0
0--------
84
Z.ai · 1,000,000 ctx · Proprietary — hosted tier, no weights of its own published
0--------
85
0--------
860--------
870--------
880--------
89
OpenAI · 1,050,000 ctx · Proprietary, API only
0--------
90
IBM · Two sizes, built on the 8B and 30B Granite 4.1 base models, each with two activated-LoRA adapters and a 470M speech encoder params · 128,000 ctx · Apache 2.0
0--------
910--------
92
Alibaba · Proprietary
0--------
93
Commons Lab (Hemmingway) · 27B params · 262,144 ctx · CC BY-NC 4.0
0--------
94
Tencent · about 30B total, about 3B active (mixture of experts, per the model name) params · Apache 2.0 (per the model card metadata)
0--------
95
Tencent · 1,048,576 ctx
0--------
96
Tencent · 770B total, 49B active (mixture of experts) params · 1,000,000 ctx
0--------
97
Bilibili (Index Team) · 35B total, 3B active (mixture of experts) params · Apache 2.0
0--------
98
Thinking Machines Lab · 594B total, 17B active (mixture of experts) params · 256,000 ctx · Closed weights at release; Apache 2.0 weights promised by the vendor
0--------
99
InternLM · 0.8B, 2B and 4B (fine-tunes of Qwen3.5-0.8B, -2B and -4B) params · Apache-2.0
0--------
100
IQuestLab · 320B total, 15B active (mixture of experts, 8 of 256 experts per token) params · 524,288 ctx · Modified MIT (commercial products must display “IQuest-Q1”)
0--------
101
TypeSafe AI
0--------
102
Jared Palmer (individual) · 0.8B, 4B and 9B variants params · 8,192 ctx · Apache 2.0 — open weights
0--------
103
Moonshot AI · 1,048,576 ctx
0--------
104
Moonshot AI · 2.8T (104B active) params · 1,048,576 ctx · Kimi K3 License (open weights, not OSI)
0--------
105
Salesforce · Post-trained from NVIDIA Nemotron 3 Super, a 120-billion-parameter open model; Koa's own parameter count is not published params · Proprietary — Salesforce holds the weights and does not release them
0--------
106
Aleph Alpha · 78B total, 3B active (mixture of experts) params · 1,048,576 ctx · Apache 2.0
0--------
107
Convai Innovations · 421M (English checkpoint, ModernBERT-large encoder); 322M multilingual checkpoint (mmBERT-base) params · 512 ctx · Apache 2.0 — weights and the Python package
0--------
108
Interfaze · 4B (a LoRA adapter on Qwen3.5-4B) params · Apache-2.0
0--------
109
Liquid AI · 24B total parameters, about 2.45B active per token (mixture of experts, 64 experts per block, top-4 routing) params · 131,072 ctx · LFM 1.0
0--------
110
Liquid AI · 1.7B params · 256,000 ctx · LFM Open License v1.0
0--------
111
Light Origins · 5.5B (Hugging Face weight count; not stated by the developer) params · Apache 2.0 — open weights
0--------
112
Paradigma · 1B (dense) params · 131,072 ctx · Apache 2.0 — open weights
0--------
113
Stable AI · 400M params · stableai-limix-non-commercial-license-v1.0 — weights published, non-commercial use only
0--------
114
Microsoft AI · Proprietary
0--------
115
Microsoft AI · Proprietary
0--------
116
Microsoft AI · Proprietary
0--------
117
JetBrains · 12B total, 2.5B active (mixture of experts) params · 131,072 ctx · Apache 2.0
0--------
118
Inception · 260,000 ctx
0--------
119
Inception · 1,000,000 ctx · Proprietary (API only)
0--------
120
Inception · 128,000 ctx · Proprietary
0--------
1210--------
122
inclusionAI · 6B params · MIT
0--------
123
inclusionAI · 6B params · MIT
0--------
124
Meta · 1,000,000 ctx · Proprietary, API only
0--------
125
Meta · 1,000,000 ctx · Proprietary
0--------
126
NaiveAI · 309B total, 15.5B active (mixture of experts) params · 1,000,000 ctx · MIT
0--------
127
NASA and IBM Research · Apache 2.0 — open weights
0--------
128
Cactus Compute · 121M (full network); deployable subnetworks from 2 to 20 layers params · Apache-2.0 — open weights
0--------
129
NVIDIA · 100M params · OpenMDW License Agreement v1.1
0--------
130
NVIDIA · 8.38B parameters params · 32,768 ctx · OpenMDW-1.1 (open weights)
0--------
131
NVIDIA · 8.38B parameters params · 32,768 ctx · OpenMDW-1.1 (open weights)
0--------
132
NVIDIA · about 30B total, about 4.3B active (mixture of experts, including the audio encoder) params · 65,536 ctx · OpenMDW-1.1 (open weights, commercial use allowed)
0--------
133
NVIDIA · 4B parameters; looped design with 8 input, 20 shared recurrent and 8 output layers params · 256,000 ctx · OpenMDW License Agreement 1.1
0--------
134
Perceptron · 36,864 ctx · Proprietary
0--------
135
Klang AI · 0.6B params · CC BY 4.0
0--------
136
Perplexity · 27B params · 262,144 ctx · Apache 2.0
0--------
137
Alibaba Cloud · 262,144 ctx · Proprietary
0--------
138
Alibaba · 7B in the visual generation component (32 single-stream DiT layers) params · Qwen Research Licence — non-commercial only; commercial use needs a separate licence
0--------
139
Alibaba · 1,000,000 ctx · Proprietary
0--------
140
Alibaba Cloud (Qwen) · 1,000,000 ctx
0--------
141
Alibaba · 1,000,000 ctx · Proprietary — weights not released
0--------
142
Alibaba Cloud · 196,608 ctx · Proprietary
0--------
143
Recraft · Proprietary
0--------
144
Ricursive Intelligence · 65B (dense) params · 1,048,576 ctx · Apache 2.0 (as tagged on Hugging Face)
0--------
145
Upstage · 35B total, 3B active params · 524,288 ctx · Proprietary
0--------
146
AWS (Strands Agents) · 2B params · Apache 2.0
0--------
147
Together AI · 4B params
0--------
148
BottleCap AI · 27.8B params · 262,144 ctx · PolyForm Small Business 1.0.0
0--------
149
Mistral AI · Four sizes: 5M, 28M, 123M and 209M parameters (patch-based transformer; the 209M model has 24 layers) params · 32,768 ctx · Modified MIT
0--------
150
Bad Theory Labs · 177B total, 6.6B active params · 262,144 ctx · Qwen Community License 1.0 — open weights
0--------
151
Baidu · 3.3B params · MIT
0--------
152
Mistral AI · Not stated (API, or self-deployment by arrangement)
0--------
153
Mistral AI · 4.4B params · Apache 2.0
0--------
154
Google DeepMind · Proprietary (Google Cloud access)
0--------
155
Tencent Youtu Lab · 5B params · Youtu-Parsing licence (custom; not intended for use within the European Union)
0--------

Score = hybrid of benchmark capability (60%), review rating (25%) and community votes (15%), renormalized by available signals. Capability is the mean of normalized benchmark results; green/red deltas compare each value to the average across benchmarked models. Official (editor-verified) and community-submitted values are kept separate - community submissions only affect rankings after editor approval. Your vote, rating and submitted stats all feed this score - sign-in required, so rankings stay hard to game.

Missing a model?

Suggest one and our editors will review and add it to the board.

Sign in to suggest a model
Launches

New AI model launches

The models that launched on TrustList this month, voted up by the community.

All AI launches
  • 1

    Liquid AI's LFM2.5-24B-A2B, released 13 October 2026: an open-weights text reasoning model with 24B parameters, about 2.45B active per token, and a 131,072-token context window, aimed at on-device agents.

    TL score
  • 2

    IBM's modular Granite model, released 14 October 2026 in 8B and 30B sizes: one endpoint for chat, scored multiple-choice decisions, safety checks and English speech-to-text, under Apache 2.0.

    TL score
  • 3
    TimestralAI model

    Mistral's first time-series forecasting models, released 15 October 2026 in four sizes from 5M to 209M parameters: open weights, probabilistic multivariate forecasts and CPU inference.

    TL score
  • 4

    NVIDIA's open-weights vision-language model, released 15 October 2026: 120B parameters with 12B active, a context of up to 1M tokens, and use in reasoning, coding, tools, documents and video.

    TL score
  • 5

    Meta's open-weights Muse model, released 15 October 2026: 816B parameters with 42B active, a 1M-token context, text, image, video and audio input, and a licence that is free below a $20M service threshold.

    TL score
  • 6

    Black Forest Labs' 5.6B open-weights image model, released 15 October 2026, for text-to-image, editing and up to 10 reference images at native 4K, with builds that run on 8 GB GPUs.

    TL score
  • 7

    NVIDIA's 4B looped language model, released 16 October 2026: one checkpoint whose shared layers can run more times for harder tasks, a 256K context and an OpenMDW licence for commercial use.

    TL score
  • 8
    GlimpseAI model

    Elorian's visual thinking model, a research preview released 16 October 2026 that reasons over images and video with tools and subagents, at $0.60 in and $3.00 out per million tokens.

    TL score